Re: radeonfb and X800 cards

[Benjamin Herrenschmidt <benh@kernel.crashing.org>]

On Thu, 2007-05-17 at 22:29 +0200, Fabio Comolli wrote:

> Is there a reason why this patch can't go upstream?

Yes. A change of that magnitude will most certainly introduce
regressions. So we can either:

 - Have it in -mm for monthes trying to iron out all of them (and we'll
miss some). And struggle also since it's hard to track subtle
regressions with very big patches.

 - Have it split in smaller bits, which make it possible to bisect in
case of problem, and thus find/correct problems much faster.

That's my main issue with it at this point. But I'm ok going for route
#1 provided that you guys are willing to help with tracking regressions
down. That also mean I need to do some serious testing on powermacs
since those rely very heavily on a working radeonfb.

Ben.